YouTube’s New Smart Tools – 2024 TubeBuddy Vision & Product Announcements

TubeBuddy AI Tools

TubeBuddy’s goal is to help creators make better content by providing deep insights and predictions. The platform uses AI to analyze existing content, understand audience engagement, and forecast the impact of potential changes. 

Armed with this information, creators can grow and increase their impact in the competitive creator ecosystem.

Existing AI-Powered TubeBuddy Tools

Title Generator:

• Functionality: Enhances video views by suggesting engaging titles.
• Impact: Demonstrates an average increase of 84% in views when used.

Suggested Shorts:

• Functionality: Analyzes long-form content to identify engaging segments, suggesting potential short clips.
• Impact: Helps creators to repurpose videos for short-form content for increased engagement (33% more comments).

Retention Analyzer:

• Functionality: Provides insights into audience retention, helping creators understand engaging segments of their videos.
• Impact: Allows creators to optimize content based on viewer engagement.

Here’s an outside perspective on the Retention Analyzer from Tim Queen.

Thumbnail Analyzer:

• Functionality: AI-driven tool for optimizing video thumbnails, offering rankings and heatmaps.
• Impact: Helps creators make informed decisions, leading to 71% more views and 84% more likes.

Those are some great tools. And some pretty impressive growth statistics, if we do say so ourselves.

So what’s next?

Upcoming TubeBuddy Features (In Beta or Development):

And we’re just getting started! TubeBuddy AI has more tools in beta mode that use developing multimodal AI technology (watch this chapter of the presentation for a lesson on what that means).

If you want to be part of developing these tools, you can sign up to participate in the TubeBuddy Beta Program.

Your Niche:

• Description: Utilizes clustering systems to identify a creator’s niche and offers insights into audience behavior and interests.
• Impact: Provides an ideation platform for discovering new video ideas based on audience preferences, including content idea generation.

Thumbnail Generation:

• Description: In early stages, this tool generates multiple thumbnail options and makes edits based on your cues.
• Impact: Promises a wide range of thumbnail options, saving you time in the creative process.

Multi-Variant Testing:

The layering of TubeBuddy AI Tools.

• Description: Work out the perfect combination of title and thumbnail to predict the potential impact of your videos.
• Impact: sky’s the limit on this one.

Additionally, McFadyen teased the expansion of TubeBuddy AI tools beyond YouTube to other social media platforms. More to come on that.

What’s Next?

The upcoming features showcase TubeBuddy’s commitment to expanding its AI capabilities, offering creators advanced tools for content ideation, creation, and optimization. These features aim to provide a more personalized and efficient experience for YouTube content creators.

And that’s just the beginning.

Transcript:

On November 30th, 2022, YouTube changed forever in what seemed like an overnight change.

Creators were producing scripts, titles, and thumbnails in a matter of seconds using tools like ChatGPT and Midjourney.

But that was only the beginning. During last month’s VidSummit conference, our EVP of Technology announced new TubeBuddy tools that promised to help creators get 71% more views, 31% more comments, and 84% more likes on their videos. And still that is not all. 

During the presentation he also discussed the fundamentals of generative AI, multimodal AI, and oddly enough, how a monkey might be the next script writer. Don’t believe me. Watch for yourself.

Alex McFadyen Presentation at VidSummit

Thank you for that wonderful intro. And yeah, I hope I can live up to the London Boy thing that I was intro’ed with, right?

So as I said, my name’s Alex McFadden and I’m the EVP of Technology over at Ben Labs. And today I’m gonna talk to you about AI. ‘Cause isn’t that what everyone’s talking about? You probably know two Ben Labs more through our product called Tube Buddy, which is an amazing tool. I hope you’re already using it. And if you’re not, there’ll be a chance at the end. But we are basically a platform for video intelligence and allowing you to get better engaged behavior with your audience.

AI and TubeBuddy

What does that actually mean? Well, it means that we let you make better decisions around your video and your content. And we do that by analyzing the content and going deeper, giving you a better understanding of the content that you’re already creating, understanding the audiences and how they’re engaging with your content. And then taking those two things, rolling to them together and predicting outcomes and impact of that content so that you can make changes in your content today that’s gonna affect how it’s gonna perform in the future. So what does embracing AI mean for creators?

Embracing AI for Growth

Well, fundamentally it means growth and impact, right? That’s why you’re all here. You wanna grow your channels, you wanna grow your impact, you wanna grow the way that you exist in the social media space. And AI can absolutely help you do that, can do it in a whole bunch of ways that you can see up there. It can help you make better thumbnails by analyzing them before they go live.

It can help you with that white blank page problem that John was talking about. How do you come up with those ideas? AI is great for that.

It can even generate entire thumbnails for you that you can use on on your channel. And then going beyond understanding your audience, where do you sit in the overall YouTube ecosystem? Where is your content really leveraged and who’s really enjoying it? And then going even deeper on your audience. Yeah, you understand which content they’re consuming.

How would it be if you can understand what products they’re buying on Amazon? Pretty cool, right? But I’m getting ahead of myself. 

Generative AI

Generative AI is so hot right now, right? I’m sure that you’ve all had a conversation about generative AI in the last few days. And basically every conversation I’m having seems to revolve around it. I mean, I might be a special case, but everyone’s talking about it. It’s part of the mainstream. These three big ones here, ChatGPT, DALLE2 and Mid Journey. They’re basically household names and you have kinda have to pinch yourself to remember these didn’t exist 18 months ago. It’s been really, really quick.

And they all have one thing in common. They all generate content with ChatGPT, you’re generating text with DALLE2 and Mid Journey. It’s generating images and it’s doing it in a way and a space and a pace that we just simply couldn’t do before.

Imagine trying to make one of those images from mid journey by hand, not only the hours and days and weeks and months and years of training to be able to use Photoshop, but just making the image itself would take hours and now you can do it in seconds. It’s, it’s kind of remarkable. 

But AI’s been around since the eighties and you know, we’ve had films about it. Terminator famously why? Why has it exploded? Why has it gone so big in the last 18 months?

Types of AI

Oh, wrong button. There we go. To be able to explain that, I need to give you a little bit of a data science lesson. Bear with me. Hopefully you’ll feel smarter at the end. What you can see here on the right hand side is an example of the two different types of ai. The one on the top is discriminative ai. This is the stuff that’s been around since the eighties.

And the way that that works is you bring the AI a whole load of data, you do a whole load of training and you can then bring it a new data point. You can bring it a new dot and you can ask it is this a red or a blue dot? And it’ll tell you the answer and it’ll get it right. And that one simple idea has spawned handwriting recognition, face recognition, chess bots that can beat the best humans. 

All of that AI stuff that was happening up until the end of last year, that was all discriminative AI. The really new stuff is the one at the bottom that’s generative. AI works in a very similar way. You give it whole load of data, you do a whole load of training.

But instead of you bringing it a new dot and saying Is this red or blue? You are now asking that model for a new blue dot, that’s the prompt. Please give me a new blue dot. And it makes one out of the data that it already understands. That’s the fundamental difference between the two different types of AI discriminative.

What sort of is this generative? Give me a new dot. But that doesn’t explain why it’s got

so big in the last 18 months. For that you need the second diagram. So that first section mod arithmetic, what that diagram is showing you is the ability for large language models, which is like the underground plumbing of ai, their ability to solve mod arithmetic. Mod arithmetic is a simple mathematical thing. It’s like take 12, seven o’clock in the afternoon, add six hours to it. 

What time is it? It’s that question fairly straightforward. What you’re seeing there along the bottom, the red dotted line is a random answer. So for a really long time these models were no better than just random numbers. And when I say a really long time, I mean it, those numbers at the bottom, that’s 10 to the 18, that’s 10 with 18 zeros after it. That’s how much training it took. 

And it was still no better than random 20 zeros, the same 22. All of a sudden it’s now starting to get the answer right. And that’s what’s really exciting here is the people who developed this didn’t know when that point was gonna happen. That tipping point, it went over it and it could now work out the answer. 

This is what was happening a lot late last year and it’s really rather remarkable. ‘Cause the other thing about large language models is you don’t know what it can do until you ask it. The question, and this is one of the reasons why OpenAI opened it up to so many people with they want to test the model to see what it could do. And people absolutely loved it. Sorry, just grabbing a drink. People loved it and people started using it in a big way.

AI Adoption

If you just go online and you look at basically any industry that uses a computer and you look for this for generative AI or AI, right now the numbers are through the roof. Developers love ai. 92% of developers are saying that they’re already using generative AI in their day-to-Day work. Teachers going all in on AI to be able to help ’em be able to deliver better classroom experiences. And you can be sure the kids are using it to help cheat on their homework. Like there is no question about that. Creators too.

You guys love it. 94.5% of you are already using AI in your daily, in your daily routines.

And what’s really interesting is when we’ve done surveys and people have done surveys on the consumers of creator content, they don’t mind. They trust it the same amount as they trust normal content that was generated in more traditional means.

And this is really important and I’ll come back to that a little later. The other reason why AI feels as though it’s so hot right now is the adoption has been like nothing else. 

This, I love this graph. It shows the amount of time it took a technology to get to 50 million users. Took the telephone 50 years, the TV 22, YouTube took two. ChatGPT did it in a month. So it’s not just that we’re all getting older, I know that I certainly am, but technology is ever increasing in pace. We’re getting new things and they’re being adopted even faster. But having all these tools is great and being able to generate all this content faster than ever before is really powerful.

Special Creator Day live stream

📅 April 23, 2024
⏰ 6 – 7:30pm PST
What’s next in creator tech? Join BENlabs CEO Ricky Ray Butler, host Judah Lee, and creators King Bach, Hannah Stocking, and Kris London for a special Creator Day event, streaming live from LA.

Get ready!

AI and the Creator Economy

But there’s an interesting thought experiment, which actually comes from a long dead French mathematician, but bear with me, it’s kind of cool. So he came up with this idea of the infinite monkey theorem, which goes like this with infinite monkeys, infinite typewriters, and infinite time. One of them is gonna write the complete works of Shakespeare. 

And what he’s trying to say there is that with enough time and with enough options, one of the ideas is gonna be good. And with generative ai, we’ve just given every single creator, every single person on the planet, an infinite number of monkeys.

The problem is that we haven’t also given them an infinite amount of judgment or creative skills to understand the content that they’re producing. Whether it’s any good. And we’ve already started to see some platforms ban generative content because they’re just getting swamped

with a huge deluge that they just can’t deal with. And I think actually going back to that slide, this is why consumers of influencer content still trust it the same amount, whether it’s been built the AI or not, because ultimately it’s your brand.

It’s you who’s putting the content out. AI is just a tool and you need to keep that in mind.

Yes, you can use these tools and please do, and I’m gonna, the rest of this pitch is all about why you should be doing that. But keep in mind it’s still your brand, it still needs to be you. And there’s a lot of you out there.

And this is one of the reasons why at Tube Buddy and Ben Labs, we think that that the creator economy specifically is going to be the forefront of this AI revolution because there’s so many of you out there who understand what good content looks like.

You have that editing skill, you have that determination, you have that ability to see what’s being created and whether it’s good and AI needs that to grow. One quick word of warning, yes there is a huge gold rush going on right now in the AI space. There are so many companies that are popping up left, right and center who are calling themselves AI startups. Some of them are doing genuinely innovative things and are fantastic.

Some of them are just kind of repackaging someone else’s stuff because they want your money and sometimes they don’t just want their money, they want your data. And I’ll come onto that in a little bit. But when you are looking at these AI twos do be a little bit careful.

There was a great talk yesterday about the risks of having your account hacked. When you share your credentials, you do need to do your due diligence ’cause there’s a lot of them out there and nine times out of 10, all they’re doing is taking someone else’s work and putting a nice new shiny layer on it and then getting you to pay for that. So one of the things you can look at, look at how many data scientists or data engineers that company employees, if the answer is zero, they’re probably not doing anything innovative and they’re just using someone else’s work.

And why are they doing that? Because of the data. The data on the entertainment space is an absolute goldmine for AI because this new generative world, it needs the unstructured data.

The Value of Data

Some of you may have heard the term unstructured and structured data before. Basically structured data is things that fit into a spreadsheet. It’s your likes, it’s your views, it’s your subscribers, it’s the countries in which the content was generated. And that’s great and you can do some really interesting stuff with that.

But generative AI needs unstructured data. It needs your comments, it needs the descriptions, it needs the videos, it needs the thumbnails. That’s what it’s all built on. And that’s why again, we think that the creator space is gonna be such a big boon for AI in general. Yeah, a picture’s worth a thousand words. What about a video? And here at Ben Labs and at Tube buddy, we’ve been doing this for quite some time. This is a, a very early alpha tool that we use internally,

but I wanna share with you ’cause it’s so awesome. What you’re seeing on, on the right there is 350,000 Instagram accounts and they’ve been grouped together not by the content that they produce, not by the country they came from or their size, it’s based on the way that the audience interacts with the content. 

So if an audience member comments on two pieces of content, it pulls it together in that cluster. And this is great. We’ve been using it to power all sorts of amazing things. And one of the things we did with it is we turned that into a map. So we now have a map of those 350,000 and I can drill in, I can go and find a cluster, a group of audience who behave the same way.

That could be they’re interested in dog walking or they’re interested in photo editing or they’re, they’re cooks for example. And this is great, I’ve spent more hours than is probably healthy digging through that map and finding the big movers and shakers in each of those audience clusters. But what gets even more powerful is when you take an individual account and you say, where is this in the map? And you can see, oh right, they’re right there. They’re in the middle of that cluster. But as well as being in the middle of that cluster, they’ve got branches out as you can see, two other clusters.

And this is really powerful as a creator. ’cause now you’re seeing, yes, you’ve got your core, but you’re also able to understand, hey, but I’ve also got an audience who’s really into dog walking. And so maybe my, my channel is mostly an IRL channel about you know, parenting, but maybe I could do some dog walking content and that’ll expand out my audience. And this is only possible because we’re not just using the structured data, we’re going into the unstructured data.

I love this slide, it’s one of my favorites. So what this is trying to show you here is that demographic data simply isn’t enough anymore.

Demographic Data vs Unstructured Data

If you were to cover up those two pictures and you just look at the demographics, it’s the same person. They’re both male, they’re both in the UK, they were born in the same year. They’ve had the same number of marriages. They are both very rich and very famous.

I think they both also live in a castle. There’s nothing you can put between them. But I would be very confident that if you’ve got their YouTube histories, you would be able to tell ’em apart almost immediately. And that’s the point, it’s the unstructured data that lets you get really deep and lets you get to really know your audience. And that’s what we’ve been doing at BEN.

We’ve been using this unstructured data along with the structured and along with our industry expertise to build models that can help creators and help brands be more effective in the space. Predicting views, predicting watch time, but even getting down and predicting clicks and conversions, it’s, it’s really amazing. And when we’ve been putting that data into Tube buddy,

the results speak for themselves. If you use Tube Buddy or if you use tools like IT and AI and integrate that into your daily workflow, you will get more subs, you will get better growth. It’s just a fact. We have the numbers here to prove it.

And what’s interesting is if you use it a bit, you’ll get a bit of benefit, that’s great. But if you really integrate it into your process and you start taking advantage of these new technologies, the results can be pretty dramatic. Alright, 

– Wait, wait, stop the video. 89% more subscribers. That’s what you can expect by using Tube, buddy. So I just wanna stop the video and tell you that we pinned a link in the comments that you can use to get a 30 day trial of TubeBuddy and get access to all of those AI features that we’ve talked about along with all the features that we’re about to talk about.

So click that link down below to try out TubeBuddy. Now back to the video. – And we’ve got case study after case study.

We’ve got all the data that back this up. You come to our our stand after this talk, you can come, we can come and show you some of those things. It’s really rather remarkable.

If you embrace this new technology, it will make your content and your YouTube experience better. But we can’t talk about AI without talking about Skynet, right? It’s good the robots are coming, they’re gonna take over the world.

A lot of people are worried that they’re coming for their job specifically. There was a great talk that was at the World Economic Forum where Richard Baldwin, you may have heard this quote,

AI is not gonna take your job but someone using AI will so that you need to start using it in order to be able to compete.

But what’s interesting and and less reported is what he said later on, which is the second quote, AI is giving more power to all workers or all creatives, but especially those who are average. And that’s a really horrible way of saying this, but I think actually what he’s trying to say there is, if you’ve got a weakness, lean on AI.

If your ability to optimize a thumbnail isn’t as good as maybe other people in your space, AI can bring that skill up. If you’ve got a a problem where you don’t know you’ve got that blank page problem you’ve got, you’re trying to idea and you can’t come up with an idea, lean on ai, it’s gonna help you. So where you’ve got a weakness, lean on AI and then focus on the things that make you special, that make you stand out. That’s the thing that you really need to be focusing on.

So now I’ve talked a lot about the theory. Finally I’m gonna get to some of the actual tools so that you know what we can do. So these are some tools that we already have live on TubeBuddy today.

Future of AI

You sign up. If you’re already a customer, you can just go in and you can use them and they’re awesome. So the first one is title generator. It was actually announced at vidcon last year, VidSummit last year.

Title Generator

And the numbers are in, if you use this tool, you will get 84% on average more views than if you don’t. It’s amazing, especially when you layer it in with something like the AB testing tool that we also have or indeed as YouTube is now starting to release, it gives you that optionality to be able to see what’s resonating with your audience with very little effort on your part.

Suggested Shorts

Suggested shorts, this one’s great as well. So what this does is it looks back through your long form content, analyzes it where you had really good engagement, really good retention on those pieces of content and then suggested those to you as little clips. You can then use that to either make shorts, maybe repurpose it, maybe make it the basis of an entire new video. It’s a great way to ideate based on real feedback from your audience. 

Retention Analyzer Tool

Also, another tool that we talked about last year was our retention analyzer tool. This actually incorporates a lot of that same technology, so you’re getting the benefit of both.

Thumbnail Analyzer

Finally my favorite, which is the thumbnail analyzer. There’s been a lot of talk of thumbnails here and rightly so, it’s the packaging. Like if everyone saw the Jon YouShaei one before, great, great talk.

What this tool allows you to do is put those thumbnails into the system ahead of time and the AI will tell you which one it thinks is gonna perform best before you’ve even put the video live. You can put in a whole bunch and as well as it just giving you a ranking, this one’s better, this one’s worse. That heat map thing you’re seeing on the other side, that’s the AI telling you which parts

of the image it thinks is most important, which ones is gonna generate the click. And so you can start making decisions about edits that you might wanna make on your thumbnail. Again, before you’ve even put it out to your customers, you get more views, you get more likes.

The numbers speak for themselves. This tool is available right now. You should all be using it if you want to get more views and more likes and more subscribers. But what I’m really here to talk to you about today is the next stuff, right?

Upcoming Features

That’s what we really wanna know. What’s the cool stuff that we’re working on that isn’t quite ready yet? The big one that you’ll hear people talking about a lot is

multimodal AI. And again, little bit of data science, sorry, but hopefully it’ll be useful. 

Multimodal AI

Multimodal AI is the idea of a AI that can use different modes of content in the same, in the same model. So that actually already exists. DaLLE2 and Midjourney are examples of multimodal AI because you give the prompt to DALLE2 in text, you say, Hey, I want a picture of the cat on the moon. And it can then make you the image. 

You can actually let, fewer people know this with some of the models, you can actually give it an image and then it will then describe it, bring it back into text. Multimodal is that idea when you have different types of content in the same model, text and images, just the start video is coming for sure.

Audio is another really interesting one. Imagine being able to do your thumbnail and be like, give me a soundscape for what this thumbnail looks like. Really, really cool. And then even three D modeling and stuff like that. There’s really a truly mind boggling amount of applications when you start being able to mix between these modalities. And so that’s what we’ve been playing with.

Everyone knows this. I think it’s been referenced many times over the last few days, the whole close your mouth thing. And he was able to do that because he has a huge team of very smart people and he has a massive brain in terms of coming up with these new YouTube thumbnail ideas.

Thumbnail Generation

Not everyone has that amount of time. Not everyone has a YouTube thumbnail idea expert, so

we’ve kind of built something to do that for you. What you’re seeing here is a model that we trained on one of our researchers, that’s him there, you can see him in the booth later. And we gave it the same prompt. We told it what good YouTube thumbnails look like.

We trained it on the person’s face so that they’re in there as well. And we gave it the prompt. I ate the world’s largest size pizza. Now this is very early alpha stuff. I’m giving you sneak peeks of stuff that’s not entirely fully baked if you like. So that’s why you’ve got some of the weird pizza shapes and things. But it gives you some ideas to work off and it can continue and it can generate hundreds of these like lots and lots of ideas for you to ideate on and view then to take forward.

They also ran it on me, which was fun. And again some weird quirks there. You can see that the one on the bottom right, for some reason it thinks I’m an octogenarian.

I think it’s ’cause my hair’s tied back. We’re gonna have to make sure it makes, it’s got like a attractiveness filter in there. Maybe we can do that. 

But I think the one on the bottom left actually is a pretty good thumbnail and could potentially do quite well as well as thumbnails. We want another big thing on the thumbnails. Yes, you gotta come up with the idea, but then editing it. So we’ve been building models that allow you to automatically edit using text.

Automated Editing

So the one on the left, it’s a nice, it’s similar a beach, but it’s a bit too rocky, maybe we wanted it at sunset. So we can give that prompt and the system is then able to automatically edit it and you can then run those as additional options.

Where it gets really exciting though is when you start layering all these models together. And that’s what we’re seeing increasingly is people picking up different AI tools, jamming them together and making more of their consistent parts. And that’s what we’ve got here.

We’re taking our thumbnail generator, we’re taking our title generator and we’re adding our click-through detector, the thumbnail analyzer and putting it all together. And what’s really great there is that we can do multi-variant testing. We can select one thumbnail and try it against all the different titles, another thumbnail against all the different titles.

So we can work out the perfect combination of title and thumbnail to present to you as a potential idea to take forward. Which is, which is gonna save you at a huge amount of time and hopefully get you those views that you are looking for.

And we’re not just looking at YouTube, we’re going beyond, yes, creators are obviously on YouTube and it’s where we started, but Instagram, TikTok, Twitch, Multi-platform Approach

all of these different platforms, the same ideas apply, it’s all unstructured data and a lot of the models that we’ve already produced, like for example the YouTube shorts, you can use that to be able to create TikTok content from your YouTube long, long form content that’s meant we’ve had to build some really new technology. 

New TubeBuddy Beta Experience

And once again, I’m gonna give you a sneak peek of some of the things that are coming out very soon with Tube Buddy. So this is the new tube buddy beater experience and it’s amazing. All you have to do is sign up and it will use that clustering system that I showed you at the beginning to find where your niche is, to find your spot in YouTube.

You can then go through and say, yeah, I like that YouTube or I don’t like that creator, or this is a really aspirational creator that I want to be more like. It can then take that and describe your audience, give you insights in terms of the things they’re searching for, the things they find interesting. And even going further.

We can sell you the sorts of things that they’re buying, for example, on YouTube, on on, on Amazon or the content that they’re reading on Reddit. And this is an ideation platform for you to be able to find new video ideas. I hope you’ll agree is pretty amazing.

But we didn’t stop there, we went further. So now you’ve got all of this data in one place.

We, we’ve got a system called Actions, which allows us to then run AI on top of the AI all over again.

AI on AI

The example that you can see there to the right is the ability to create new ideas. I dunno if some of you have played with Chat GPT or some of the other title generator or similar tools.

Yes they work, they can give you some good ideas, but they’re kind of limited. They end up being a bit generic quite quickly and maybe they’re not really clued in on exactly who you are as a creator. By us getting all this data to begin with, we can create some really targeted, really specific ideas for you just for your channel.

And when we’ve been early alpha testing this, the results have been, have been, have been amazing. There was one that I particularly remember. We were going through the demo and the gentleman that we were showing it to stopped us and said, wait a minute, have you got access to my email?

And we’re like, no. Why? Why do you think we’ve got access to your email? Because that third option that you’ve got right there, I’m doing right now, like it’s in the final stages of post-production. That’s how good these predictions have got.

We dug through deeper and he said, oh stop a second. I wanna go and phone someone. Why? Oh, ’cause the AI has just suggested that I work with someone that I was, did a couple of collaborations back when I first started my YouTube career and I haven’t spoken to them in years.

That sort of human connection, that ability to spark creativity is is absolutely priceless.

And it’s just so great to see the technology that we’re building, making a real difference in creator’s lives. So now the important bit, some QR codes, I understand the internet here is rubbish, so maybe take a picture and try and do it later.

Conclusion and Q&A

But that first one that will get you 90 days worth of TubeBuddy Legend and access to all of those AI tools that are already live completely free of charge. But most importantly on that, on that page, there’s a section there that you can see that says, I’m interested in joining the beta program.

If you tick that on, you are then liable for inclusion in that second piece that I was showing you all the really, really cutting edge stuff. So make sure you tick that on ’cause then we can give you access to that. If you want some demos of some of these tools a little bit later, we have a booth, come and see us. We’d be more than happy to show you some of the really cool tech that we’re working on.

And finally, if you wanna talk nerdy, I’m on LinkedIn, that’s me. You can hit me up and I’ll, I’ll happily talk about AI till the cows come home. And on that, thank you very much. I hope you found it at least a little bit interesting, a little bit insightful and I’ll happily take some questions. 

Q: Hello. So I’ve been really interested in the new things that AI can enable. So I’ve been trying to make scripted short films and I want to use Mid Journey in order to generate images

and be able to tell whatever stories I want. So I, I’m just really interested in this space and like what you think of scripted content and how AI is gonna enable new things to be created that no one else could create before.

A: Absolutely. So I think there’s all sorts of interesting opportunities there. One of the other speakers I was say I was talking to was saying that they use things like ChatGPT

and Midjourney for Ideation. So they’ll write down their initial script, they’ll then take it to ChatGPT and say, condense this, make it funnier, make it shorter, make it more punchy. 

They’ll then use that, then they can go to Midjourney and use that as a storyboarding tool. Take a couple of those ideas and be like, Hey, what would this scene look like? Visualize this for me.

And those are exactly the sorts of tools that we’re building into Tube Buddy. So yeah, the future is here. It’s just not “evenly distributed” yet. One of my favorite quotes from one of my favorite authors, thank you, thank you.

Q: Thank you for your presentation. Thanks for TubeBuddy. It’s amazing. Thank you. Maybe this is just my humanity and existential crisis. Yes. But I worry about using AI and then kind of getting blocked because Google or YouTube or whatever yeah. Thinks like, hey, this is an organic content and we want to prioritize organic content. Yeah. So can you speak to that and assuage my fears? Yes. 

A: Yeah, I’ll try. I’ll try. So it’s an interesting problem and yeah, there are, there are AI companies that are trying to work on AI detection and things like that for exactly the reason that you pose. Amazon actually recently have limited the number of self-published books that they’re accepting

because they were getting completely swamped with AI generated content. I think what it comes back to is that one of those earlier slides, I’m not gonna try and go through ’cause it’ll go wrong, but where it was like the, the viewers of your content trust, your AI generated content or AI assisted content as much as your primary content. 

No one cares whether you used Avid or Final Cut Pro. No one cares. Whether you use Photoshop or some other tool for your thumbnail, AI is just another tool. Ultimately it’s your expression. That’s what they’re following. They’re following you because you are unique and you have your own perspective on the world. I don’t see these platforms limiting the use of these technologies except in the cases where they’re being used to generate mass content that has very low value. Fundamentally, if AI helps you make better content that’s in the platform’s interest.

Q: Hello? Oh, there we go. I just got a quick question. Maybe simple, Jon, I thought he had a great idea as far as using shorts to determine the intro with the short suggestion. Is there a way to upload it directly to TubeBuddy prior to publishing it?

A: So the, the short Suggestion tool uses your existing long form content and your audience’s reaction to it. So that’s after the fact. Whereas what John was showing you was the idea of doing it ahead of time to be able to tease out and work out which things might work. Okay. 

Where I think short, the short suggestion tool is gonna be really good is then using that, learning what your audience starts to like and then using that in your future videos. Or indeed, looking back at some of your really old videos, what really resonated there? And then maybe using that as that short little clip as the basis for an entirely new piece of long form content.

Q: So you sort of touched on this two questions or two answers ago, but legalities of ai Yes.

If we’re inside of TubeBuddy, I, I get it, you know, we’re covered by you guys, but if it’s outside of TubeBuddy using an ai, we have an in-house legal team that told our marketing crew, it’s like, don’t be touching anything with AI right now.

A: Yeah. It’s, it’s certainly, it’s, it is an evolving space.

The the, the, the legal landscape, just the, the sheer number of companies that are out there that are doing stuff is, is mind boggling. Again, I think you just have to look at the, there is some legal sides of it in terms of what the models were trained on and whether they’re allowed to even to be used for commercial use. Facebook and things like that have been releasing a lot of their models as open source available for commercial use.

And so that’s great. But yeah, this is why when you are using some of these new tools, you do need to look at it a little bit closely to see what rights you have over the content, what rights they’re retaining, and indeed where their content was trained from.

So I don’t have a short answer for you, unfortunately, it’s a very nuanced and difficult question, but it is one that even the EU right now, there is a huge new law coming through in the EU called the EU AI Act that is trying to address some of the potential harms and potential legal ramifications of ai. But this is, it was, didn’t exist 18 months ago.

Everyone’s scrambling to keep up. So pay attention, keep online use Tube buddy. ’cause we’re, we are there to help and support you. But yeah, it’s a, it’s a difficult space for sure. 

– Last question is from Big Mike

Q: Howdy there, I do all source of content on Route 66 and my demographic is a much older demographic, 45 to 65 plus. And I’m wanting to find out, is there a way I just got set up with your legacy program, but I haven’t had a chance to play with it much. Okay. Is there ways of putting demographics in there because a 20 year old’s probably looking at different likes it faster, likes it different things. Is that something you guys, is it available now or is that something you guys are working on? 

A: So it’s an interesting point and yeah, when you build, when you train ais, you wanna be able to expose it to as much data as possible. But one of the key things there is I, yes, that may be the skew of your demographic, that it may mostly be older, but if you limit it only to that demographic, your younger audience that may still enjoy it, you would be excluding them.

That’s why the audience clustering system is so powerful because it doesn’t care what the demographic is. It cares whether they like your content. That’s what it’s focusing in on. It’s your actual niche. You are much more varied and much more multifaceted than a really limited demographic profile. 

And that’s one of the things that’s so powerful with the unstructured data of generative AI is it allows you to keep that nuance alive without having to condense it down to you are a millennial, you are a boomer. Those are such broad brushes that you can’t build the level of engagement that you want with your audiences when you’re using such a crude tool.

Special Creator Day live stream

📅 April 23, 2024
⏰ 6 – 7:30pm PST
What’s next in creator tech? Join BENlabs CEO Ricky Ray Butler, host Judah Lee, and creators King Bach, Hannah Stocking, and Kris London for a special Creator Day event, streaming live from LA.

Get ready!